Agilent E3631A Power Supply humidity damage

[Alex]

I have a 3631A which was sat in high humidity for a few months.

On initial power up, the measurements on the screen were all over the shop. Now the screen lights up with the test pattern and it will then either switch off or display ERROR.

The rotary encoder works with audible clicks from the buzzer inside, there is no voltage on the outputs.

Running the diagnostic takes a long time and says FAIL but no error code.

The fan is running. I have started probing around 'Board B' the +/-25V reference board and the +/- 15V supplies are OK. On the DAC circuit, the DAC (U12) is powered OK and its reference is OK. The buffered references for CV and CC do not appear to change as I try to tweak the relevant outputs.

I suspect the damage is due to dust collecting water.

Has anyone come across this before? I will continue probing.

A

[Noise Floor]

How high is "high"?

[Alex]

~93%

[Noise Floor]

Hmm.  How long did you let it "dry out" before powering on? 
Based on what you are getting its hard to offer much help, at least for me.

[lostboy]

When I repaired mine I found the opto isolators were very sensitive to contamination. A thorough clean in stages - with a final rinse of clean IPA of both boards might be a good start - it's difficult to get contamination out from under the chips so  you may have to be prepared to lift and replace in some cases. Pay attention in and around the chip leads.

Chris

[Alex]

I am told this was just powered up without a chance to dry. 

I will proceed by stripping down and thoroughly cleaning. Then I will systematically test power rails although I suspect the damage is to sensitive circuitry. The fact that the screen switches on then off is intriguing.

I've heard this humidity problem before with ATX power supplies, they too collect plenty dust inside.

A

[Alex]

I have now completed repairs to this power supply.

Several components including resistors were damaged, primarily in the discrete ADC section. Other faulty components found were the CC/CV comparator, 6V section pass element driver op-amp, analogue isolators and the digital isolators for the CC/CV and RS232 signals.

I am attaching a picture of all that was replaced.

A

[CaptainObvious]

Just out of curiosity (and to learn something, I suppose;)):

How did you go about determining this parts in particular had failed? Were they visibly damaged or did you actually go through and start testing individual parts of the circuit? (Or after finding one bad part, you just jump the gun and replaced them all?)

Either way, congrats on your repair!

[Alex]

Hi CO,

Primarily by breaking down the problem and solving it one part at a time.

Having a general knowledge of the architecture of a common power supply helped me focus on the system's functional parts that are most likely to be affected. For example, when I was switching it on the displayed I/V was fluctuating randomly, occasionally displaying values outside the specs of this power supply. This implies that there is something wrong in the ADC or power section.

I could further break down the problem using the full schematics of the unit found online. Then you can probe individual subcircuits (starting with voltages is helpful) and with your electronics knowledge make a call if they are faulty. For example, an inverting op-amp topology was outputting its rail voltage while R1=R2. That's not right, and it turned out one of the resistors was open-circuit leading to no feedback.

Doing some research online I read that the analogue optocouplers are quite sensitive and prone to failure. I checked the voltage across the LED and it was 2.4V when you can see in their datasheet that the typical value is 1.4V. So these were open-circuit.

The heart of the ADC section is current-based, so probing with a voltmeter gave little information. If I suspected something was faulty, I made a call to replace it as this is easier than running more extensive diagnostics. There was a buffer op-amp outputting something different to its input while driving a reasonably high impedance, so that op-amp was faulty.

When the supply started working again I did a short-circuit test and from CV I was expecting it to switch to CC. It displayed 'Unreg', suggesting the CC/CV comparator was faulty and gave the wrong signals to the main processor.

In the end everything worked OK but it still displayed ERROR. I was about to just leave it as it is, thinking that the fault was around the RS232/GPIB section which I don't need. I probed one of the optocouplers as I had to replace a similar one and voila, it was faulty.

Overall I was lucky that the main processor (integral part of the ADC) was not damaged, I would not be able to replace that easily.

This would have not been a commercially viable repair, a lot of time and money was spent on finding and replacing components. But it is a good power supply to have on the bench.

[Stray Electron]

  Did all of those parts prove to be faulty?  If so, that's a LOT of failures in one device!

[Alex]

Hi Stray, with the exception of one precision op-amp, the rest are confirmed faulty by in-circuit measurements before I replaced them.

A